General Removal by the Numbers

Getting removal right always seems like one of the hardest aspects of cube design. Ars Arcanum is a stat-based limited article series we have discussed here before. Today he came out with a detailed analysis of removal across a large portion of Magic's history. The results should be immeasurably valuable in helping craft cube environments with removal that feels right.

http://puremtgo.com/articles/ars-arcanum-history-removal

Here's a sample chart showing the amount of removal in each set:
xqpyTlH.png


It's a long read with lots of detail, but I thought it was fascinating. In particular, he demonstrated a near constant trend in the total amount of removal (~20% of packs), but overall that removal has been getting worse in a variety of important ways. Here is his summary at the end of the article:

Here is a summary of what we saw:
1. The amount of removal per pack has stayed essentially the same since Invasion.
2. The converted mana cost of removal spells has steadily increased, though not by as much as I assumed.
3. Removal spells deal more damage now than in the past, but this has not scaled proportionally with the power level of creatures.
4. Removal spells are harder to use and are less versatile than they have ever been.
5. They are also less effective than ever, and just don’t kill things dead.
6. The most significant change is that removal no longer gives you much additional value beyond just killing a creature.

I just ran the numbers on my cube's removal: 19.3%. That is actually pretty encouraging and explains some of the "retail feel" comments I have received from players.

So what does this mean for our removal packages?
 

Grillo_Parlante

Contributor
This is really weird. He has 20% removal as the average, which he says is an as-fan of 3 per pack. He also says that every player will get about 9 removal spells on average per draft (which is a large number) but concedes that only 3-5 of them will be good removal spells.

I think when we had this discussion before, an as-fan of 3 was high, while an as-fan of 2 was low.
 

FlowerSunRain

Contributor
He's counting all kinds of stuff as removal, including unsummons, sleep effects, counterspells and tappers which are arguably not removal.
 

Eric Chan

Hyalopterous Lemure
Staff member
I suspect the difference is that every single piece of removal that we run as cube designers is intended to be viable, and not embarrassing to maindeck; whereas in retail limited formats, many of the removal spells are unwieldy by design, and only borderline maindeckable.

We'd probably also need to go into our own cubes and look at the average converted mana cost, the restrictiveness, and the effectiveness (to use three of the author's other metrics) of our own removal suite and see how that lines up, before we can draw any conclusions.
 
His definition of removal as a category is pretty well-aligned with my current removal flagging as well. I plan to go through his categorization system with my removal and see how the numbers fall out. What I would give for a copy of his raw data...
 

Onderzeeboot

Ecstatic Orb
He's counting all kinds of stuff as removal, including unsummons, sleep effects, counterspells and tappers which are arguably not removal.
I think these are all very valid spells to count as removal, considering what he's trying to measure. Removal can be permanent, but it can also be temporary.
 
Cool article. Thanks for posting that.

I agree in particular with the card advantage trend. I don't really like CA strapped to removal spells because it is so insanely efficient that it can (in extreme cases) nullify all but the spikiest of strategies which just kills a lot of creative deck ideas. I still run a lot of those CA removal type effects (not going to lie here), but I have started culling many of them and only keeping the more conditional ones (things like bone shredder where you have both a fairly weak body and echo to pay for your CA). Flametongue Kavuis probably the poster child for this kind of effect and maybe I should just kill that sacred cow and let it go (say for Pia and Karin in the upcoming set), but it's one of those historical cards like Hippie that I just can't bring myself to cut even if all the data says I should. Part of what I enjoy in cube is the nostalgia and there are a few cards like this which give me a warm fuzzy.
 
I suspect the difference is that every single piece of removal that we run as cube designers is intended to be viable, and not embarrassing to maindeck; whereas in retail limited formats, many of the removal spells are unwieldy by design, and only borderline maindeckable.

We'd probably also need to go into our own cubes and look at the average converted mana cost, the restrictiveness, and the effectiveness (to use three of the author's other metrics) of our own removal suite and see how that lines up, before we can draw any conclusions.


While this article was interesting, I think the data becomes less useful the more powerful your environment is. While it's true that many removal spells in retail formats are crap and borderline unplayable, so are many of the other cards - you have to "make it work" with what you've got. If your cube is lower powered, mirroring the efficacy of a retail set makes sense - no one is going to be happy if you've got a playset of Lightning Bolts and Arc Lightnings for removal in your cube if T2 Bosk Banneret into T3 Avian Changeling and War Falcon are the explosive plays that define your format.

Cube is an entirely different animal than retail, so we have to acknowledge that. There is very little filler - only enough to support archetypes that justify it, really - so our removal should, as well, be better. If you're using crap removal in limited to remove a bomb, that's fine - if you're using crap removal to remove one of the many good cards your opponent is using in cube, you aren't doing so fine anymore. It's perfectly okay if you want a creature-heavy environment and have very little removal or very weak removal as a result - that's a deliberate choice you're entitled to make. But don't take cues from retail data without a grain of salt, or, perhaps more appropriately, a few heaping Salt Flats.

That said, I tightly control my removal suite, both for flavour reasons and to keep the format healthy, so I was happy to find that, based on my own measuring metrics, my removal clocks in at 19.67%, which is similar to a retail set and gives me a fairly mean AS-FAN of about 2.95 pieces of removal per pack, which is great by me, given the higher variance of my cube and the managed power level of those removal tools.
 

Grillo_Parlante

Contributor
I guess we'll have to wait until his follow-up article, as I have a bunch of questions.

One thing that seems strange to me (though perhaps is a necessary evil) is that he dosen't take into account the nature of the format when assessing effectiveness. Silent departure was first pickable in III, but in that format unsummon was really good. He hints at this a little bit when he talks about how ineffective pacifism effects are against utility creatures--but what if the format runs lots of pacificisms but few utility creatures? Counterspells are also pretty bad in a lot of limited formats.

There is also a really sweet conversation to have about limitations. We've talked a little bit in the past about encouraging diversified answer suites for different threats. The sorts of limitations you see in retail formats lately seem to be largely imbalanced given the power level of the bombs, but I like stuff like terror clauses. Tying this back with the latter point about format architecture, bone shredder is pretty reasonable as a source of card advantage removal in my artifact heavy cube, precisely due to its limitations. I imagine the same could be said of flametongue kavu in a format with high toughness (ROE).

But yes, there is a lot about his metric that has to be adjusted for cube. And this opens us up to a critical discussion of how removal should be catagorized in a cube format, as otherwise his data is hard to use.
 
I'm not necessarily in favor of bad removal per se, just suggesting that a lot of cubes would be better without cards like StP.
 

Onderzeeboot

Ecstatic Orb
I actually cut Flametongue Kavu a while ago, though maybe it's just the card that make red removal not suck. He does mention red is the only color that got stuck in 1999 (well ok, I think he said 2000) with the power of its removal.
 

Onderzeeboot

Ecstatic Orb
I am in the process of adding Origins to my new cube, and want to cut some cards to end up at 450 again. When asking myself if I could conceivable cut a removal spell, I stumbled upon this article by Alex Ullman. He counts only those cards able to remove a creature for more than one turn. Turns out I'm at an as-fan of 3.75. I think it's time to cut some removal!
 
Top